The abundance of the 40 commonest OTUs in the River Wensum accounted for approximately 42% of the total abundance in the river water. This percentage is lower than that obtained by ARISA where the 40 commonest DNA fragment sizes accounted for about 55% of the total bacterial abundance, indicating that ARISA gave higher abundance but lower diversity than 454 pyrosequencing. Zhu et al. (2013) investigated bacterial community composition in sea sediments in China using pyrosequencing. In shallow sea sediments, 62 abundant OTUs made up 22% of sequences, while 62 common OTUs in deep sea sediments made up about 57% of total sequences. In seawater samples, Chow et al. (2013) found that the five most abundant OTUs from different depths (0 to 5 m) made up about 52% of the total bacterial sequences. Hence, the samples from the River Wensum are less diverse than shallow water sediments (Zhu et al. 2013) , but show similar diversity to deep sea sediments (Zhu et al. 2013) and also seawater samples (Chow et al. 2013).
The abundance of the 40 commonest OTUs increases as water moves downstream in the River Wensum and the highest abundance of all these commonest OTUs was found at sites SC and S18. The downstream site S18 is at the outflow of the study catchment, a 4th order stream located in an urban area. This location receives organic matter and bacteria from upstream sites and as runoff from urban areas and sewage treatment works. Site SC also shows a high abundance of all the commonest OTUs especially Cyanobacteria, despite being an upstream site. These organisms are growing at a site with presumably preferable environmental conditions. Increased Cyanobacteria at site SC may due to discharges from septic sewage (Ahmed et al. 2005) or an adjacent lake which is connected to the stream and can release these bacteria into it. In arctic tundra, Crump et al. (2007) found that the bacterial community composition in swage runoff was very similar to connected lakes, and was attributed to dispersal processes. Also Nelson et al. (2009) found matching results for different lakes and streams in California, USA.
The commonest bacteria at upstream sites were Proteobacteria (Beta, Delta and Gamma) and Bacteroidetes (Sphingobacteria and Flavobacteria). On the other hand, Cyanobacteria, Actinobacteria, Bacteroidetes (Cytophagia and Flavobacteria) and Alphaproteobacteria were the commonest bacteria at the downstream sites.
158 Sekiguchi et al. (2002) determined the succession of bacterial community structure along the River Changjiang in China. Upstream sites were dominated by Betaproteobacteria and Bacteroidetes. These decreased as water moved to downstream sites of the river with bacterial communities becoming more dominated by gram-positive bacteria, such as Actinobacteria. Sekiguchi et al. (2002) attributed the succession to changes of nutrients, water temperature, river flow and pH. In different streams in Spain, Simek et al. (2001) found that Betaproteobacteria and Cytophaga/Flavobacterium were the most common bacteria at upstream sites and were largely allochthonous in origin.
This study showed that the most common organisms at upstream sites are soil bacteria (OTUs 2, 29, 32 and 9), indicating the importance of allochthonous bacteria. These comprise freshwater bacteria (OTUs 12 and 23) and one sewage bacteria (OTU4). These organisms play different roles in these environments. For example, Betaproteobacteria (OTU2) have a role in Fe (III)-reduction (Finneran et al. 2003), Betaproteobacteria (OTU4) in the degradation of various pollutants and dissolved organic matter (Hiraishi et al. 1997), Bacteroidetes (OTU9) in the degradation of nutrients (Weon et al. 2009) and Betaproteobacteria (OTU17) in the reduction of organic sulfur (Bruland et al. 2009). Bacteroidetes (OTU23) catalyse organic matter in cold environments (Nogi et al. 2005), Deltaproteobacteria (OUT29) prey on the other gram-negative bacteria (Pineiro et al. 2008) and Gammaproteobacteria (OTU32) degrade cellulosic plant fibres (Lednicka et al. 2000).
The majority of the commonest bacteria in the downstream section of the River Wensum are freshwater bacteria (OTUs 1, 3, 10, 19 and 26). Two of the OTUs (10 and 20) are soil bacteria. These organisms play an important role in these environments. For example, Cyanobacteria (OTU1) in N2-fixation (photosynthetic bacteria) (Welsh et al. 2008), Bacteroidetes (OTU19) in reducing nitrate (Weeks 1954) and Alphaproteobacteria (OTU26) in being phototrophic, photoheterotrophic and chemoheterotrophic (Arunasri et al. 2008).
In December 2012, during high river flow events (mean; 9.6 m3/s at Costessey Mill) and rainfall (82 mm), the commonest bacteria found were Betaproteobacteria (OUT 2, 17, 34 and 40) followed by Alphaproteobacteria (OTU26). Chen et al. (2013) assessed the effects of dry and wet seasons on bacterial structure in different streams along the River Chongqing in China and found that bacterial diversity and abundance were greater in the wet season compared to the dry season with Betaproteobacteria more common in the wet season and Actinobacteria in the dry season. Changes in bacterial communities show most for Betaproteobacteria in the wet season (high rainfall events), which are not only attributed to environmental parameters, such as water temperature, but also to terrestrial source areas and tributaries that discharge into stream waters.
This study found that high rainfall and high flow events during December 2012 discharged many of the commonest OTUs from terrestrial source area and groundwater into upstream sites of the River Wensum (SA, SB, SC, SD, SE, SF and S20), increasing these commonest bacteria. At upstream sites on the Mississippi River, abundance and diversity of most common members of the dominant bacterial groups were also positively related to rainfall (Staley et al. 2013) as
159 a result of rainfall carrying bacteria from terrestrial source areas into streams, and so affecting the relative abundance of these bacterial communities rather than their presence or absence. In the research carried out in this study, 454 pyrosequencing and ARISA (Chapter 4) were found to be adequate in assessing the bacterial community composition present of the River Wensum. Unlike ARISA, 454 pyrosequencing makes it possible to identify actual microorganisms. However, 454 pyrosequencing is relatively expensive and prohibits processing high numbers of samples.
For the future, Illumina sequencing, which can give potentially 109 500 bp reads offers the benefits of both 454 pyrosequencing and ARISA. It is possible to have barcodes for both the forward and reverse primers. Hence, with 96 of each, it is possible to multiplex 96 × 96 = 9216 samples (Harbers and Kahl 2012; Liu et al. 2012; Wong et al. 2013). Such that, for a very large study, the costs are competitive with ARISA.
5.6 Summary
In the research presented in this chapter, 454 pyrosequencing offered an insight into the bacterial community composition of the River Wensum. The dominant bacterial groups in the river water were found to be common in various freshwater environments and were Proteobacteria, Bacteroidetes, Cyanobacteria and Actinobacteria. The taxonomic affinities of these common bacteria provide important information about their relatives, the first environments from which they were isolated and also their environmental roles. For example, most members in the River Wensum belong to the commonest bacterial OTUs that are involved in the cycling of nitrogen and metals. The abundance of the majority of the commonest bacteria were found to increase as water moves downstream (3rd-4th stream order), with the highest abundance recorded at sites S18 and SC. The most common bacteria at upstream sites were Proteobacteria (Beta, Gamma and Delta) and Bacteroidetes (Sphingobacteria and Flavobacteria). The majority of Proteobacteria and Bacteroidetes at upstream sites are considered to be soil bacteria and these decreased in abundance as water moved downstream. The commonest bacteria at downstream sites in the River Wensum are Cyanobacteria that are involved in N2-fixation, Bacteroidetes (Cytophagia and Flavobacteria) and Alphaproteobacteria. Most of these bacteria are considered to be freshwater bacteria and increased in abundance at downstream sites. The common bacteria in December 2012 were Betaproteobacteria and Alphaproteobacteria. These bacteria are soil bacteria that can be flushed from terrestrial source areas into streams after flood events. The relative abundance of the commonest OTUs changed in December 2012 compared to February 2012, suggesting the role of high rainfall and flow rate at 10 selected sites in shifting the relative abundance of bacterial communities between these two periods.
160 Chapter Six
Conclusions and future work recommendations
6.1 Conclusions
Determining bacterial community composition and dynamics is a fundamental task in microbial ecology because of their rapid response to natural and anthropogenic pressures, and the role that they play in nutrient and carbon cycles (Kirchman et al. 2003; Daims and Wagner 2007). Most studies have been focused on marine and soil environments with much less effort on freshwater systems, despite the widespread occurrence and importance to humans of these systems worldwide (Debroas et al. 2009). Most studies assessing bacteria in freshwater environments have been focused on indicator bacteria, but composition and dynamics of bacterial communities in these environments have had less attention (Sigua et al. 2010). The effects of spatial and temporal variations, abiotic and biotic factors on bacterial community composition, structure and dynamics are still poorly understood (Lawrence et al. 2004; Lindstrom et al. 2005). Molecular techniques based on DNA have revolutionized our understanding of bacterial community composition and abundance in natural environments (Muyzer et al. 1993). In addition, the use of metagenomic approaches and other applications of high throughput sequencing methods to investigate bacterial communities in freshwaters is still rare (Debroas et al. 2009).
This study set out to investigate bacterial community composition and abundance in the River Wensum from June 2011 to February 2013. It also aimed to determine the effects of spatial and temporal variation and environmental factors on bacterial community composition and abundance. The River Wensum has been subject to inputs of high amounts of nitrogen and phosphorus as a result of intensive agriculture practices (upstream) and also discharges from sewage treatment works (downstream), causing problems for the river ecology and particularly in altering bacterial community composition and abundance. The research presented in this thesis presents one of only few studies of the bacterial community composition and abundance in a lowland arable catchment. It is also one of very few studies to carry out a detailed investigation of the temporal characterisation of bacterial communities in a river system. In addition, the microbiological techniques used here were applied for the first time to samples
161 from the river Wensum. The main conclusions from the research presented in this thesis are detailed as follows.
Total bacterial numbers were assessed from June 2011 to February 2013 using standard methods, epifluorescence microscopy and DAPI staining. It is concluded that total bacterial numbers (Chapter 3) ranged from 0.21 × 106 cells/mL to 5.34 × 106 cells/mL (mean = 1.1 × 106 cells/mL). Total bacterial numbers varied both spatially and temporally with greater differences between times than sites. Bacterial numbers increased as water moves downstream with the highest numbers recorded in 4th order streams. Temporally, the highest total bacterial numbers were recorded in June and August 2011 (summer), while the lowest numbers were recorded in December 2012 and February 2013 (winter). The variations of total bacterial numbers showed some relationship with environmental parameters including water temperature, TP, TC, TN, stream order, river flow and the numbers of adjacent sewage treatment works. Approximately 52% of the differences in total bacterial abundance were related to these parameters.
In February 2013, total heterotrophic bacteria were determined using the traditional heterotrophic bacterial count method and then the proportions of total bacterial numbers were calculated. Heterotrophic plate counts (Chapter 3) showed significant variations between sites, but did not show significant relationships to any environmental parameters. However, the highest heterotrophic plate counts were of the downstream sites S14 and S8. These two sites are downstream of the three sewage treatment works, most likely indicating the role of treated sewage in increasing heterotrophic bacterial numbers, especially those responsible for the degradation of ammonium (ammonium-oxidizing bacteria) as shown in the 454 pyrosequencing data in Chapter 5. In addition, percentages of total bacteria that are culturable ranged from 0.48% to 7%, and were negatively related to total bacterial numbers.
Shifts of bacterial community composition were assessed using the automated ribosomal intergenic analysis (ARISA) technique from June 2011 to December 2012. ARISA fingerprints (Chapter 4) showed significant spatial and temporal shifts in the composition and abundance of the bacterial community. Bacterial diversity is highest at upstream sites, while it decreases as water moves downstream. On the other hand, bacterial abundance increases as water moves downstream. However, site SC, which is impacted by septic waste discharges, is upstream and does not fit this pattern. It is more like the downstream sites and presented a high abundance of a commonly identified OTU (702.09). Multidimentional scaling (MDS) displays differences of bacterial community composition between sites and times. There is a large shift between upstream sites of the sub-catchment areas (SA, SB and SE) and downstream sites (S8 and S18) of the river Wensum and this was attributed to catchment characteristics. Upstream sites are small streams located in intensive agricultural areas, and because they are the primary link between terrestrial areas and aquatic environments, these locations receive bacteria and nutrient runoff, especially during high rainfall and flooding events. Downstream sites, on the other hand, are large streams located near urban areas and are influenced by a number of sewage treatment works (STWs). The discharge of bacteria and nutrients from both urban areas and STWs affect bacterial communities in river water.
162 Temporally, there is a large shift of bacterial composition in the same months of different years, particularly between December 2011 and December 2012. The mean water temperature was the same in both, but water flow was very different (mean= 2.30 m3/s and 9.60 m3/s respectively). Comparing September 2011 and September 2012 (mean water temperatures were 14.2 ˚C and 11 ˚C and mean water flow was 1.88 m3/s and 2.32 m3/s, respectively). Hierarchical partitioning analysis showed that water flow and temperature were the strongest factors affected the temporal variations of bacterial community composition in the River Wensum. However, bacterial community composition was very similar between summer 2011 and summer 2012, and between autumn 2011 (October and November) and autumn 2012 (October and November). Temperature and river flow therefore play an important role on the temporal variation of bacterial composition.
To investigate the significant effects of spatial and temporal variations and environmental parameters on the abundance of common bacterial OTUs in the river, the abundance of the commonest 20 OTUs were analysed individually and showed significant differences between sites and months. Common OTUs made up 40.4% of the total community. The abundance of the majority of these OTUs showed variations, with greater abundance between sites than between months in the downstream sites of the river, especially at sites S8 and S18. This result indicates that these OTUs grow and multiply in the river and that changes in their abundance are based on the fluctuations of physical and chemical parameters of the river water. On the other hand, a few OTUs exhibited greater and often increased abundance during high rainfall and flood events, suggesting that these represent bacteria of terrestrial origin that are flushed into the lower streams.
The variations in abundance of the commonest OTUs were related to a number of environmental parameters, including stream order, temperature, TN, TP, TC, TSS, pH, arable land, improved grassland, other grassland, urban area, flow rate, STWs and rainfall. These parameters accounted for different proportions of the variations of the commonest OTUs, with highest values explained for the first five commonest OTUs (16% to 45%). Changes in bacterial diversity were related to fewer environmental parameters, including TC, TN, temperature and stream order. These parameters explained about 18% of the variation of bacterial diversity. The structure of bacterial communities in the river Wensum water was determined using 454 pyrosequencing for February (26 samples) and December 2012 (10 samples). 454 pyrosequencing of 16S rRNA (Chapter 5) showed that bacterial communities in the river Wensum contained phyla that are found to be common in the other freshwater environments worldwide (Jordaan and Bezuidenhout 2013; Moller et al. 2013; Staley et al. 2013). The dominant bacterial phyla were Proteobacteria (the classes of Beta, Epsilon, Gamma, Delta and Alpha in decreasing order of importance), Bacteroidetes (the classes of Flavobacteria, Cytophagia and Sphingobacteria in decreasing order of importance), Cyanobacteria and Actinobacteria. Principal component analysis showed that the 40 commonest OTUs belonging to these dominant phyla made up 42% of individuals.
The commonest bacteria at upstream sites were Proteobacteria (OTUs 2 and 4), Deltaproteobacteria (OTU29), Gammaproteobacteria (OTU32), Sphingobacteria (OUT9) and
163 Flavobacteria (OTUs 12 and 23). Most of them (OTUs 2, 9, 17, 29 and 32) are soil bacteria, suggesting that these bacteria are terrestrial in origin and are flushed into the lower order streams. Most of them showed positive relationships with TN and TC and the presence of arable areas. On the other hand, the commonest bacteria at downstream sites were Cyanobacteria (OTU1), Flavobacteria (OTUs 3, 10 and 19), Cytophagia (OTU14), Actinobacteria (OTUs 20, 21 and 25) and Alphaproteobacteria (OTU26). Most of these bacteria are freshwater bacteria. These bacteria become more common as water moves downstream, but there is one site (SC) which is upstream that was an exception to this pattern. Bacteria are actively growing in the river, diluting other taxa and reducing the diversity as water moves downstream. Most of the bacterial OTUs showed a positive relationship with TP and the presence of urban areas. The highest abundance of all the commonest OTUs was recorded at site S18 (4th order stream) and site SC (upstream site).
Commonest bacteria in December 2012 were Betaproteobacteria (OTUs 2, 17, 34 and 40) and Alphaproteobacteria (OTU26). These are soil bacteria and showed greater abundance during this time of high rainfall and flood events, suggesting that these represent bacteria of terrestrial origin that are flushed into the lower stream order.
The majority of the 40 commonest bacterial OTUs belonging to the dominant phyla were found to be responsible for recycling of nitrogen (OTUs 1, 6, 7, 13, 15, 19 and 37), confirming that the river Wensum is exposed to high concentration of nutrients, such as nitrogen as a result of agriculture practices and discharges from sewage treatment works. Some of the bacterial OTUs are involved in Fe-reduction (OTUs 2 and 22), sulphur-reduction (OTU17), phosphorus removal (OTU5), carbon cycling (OTU31), removal of some chemical compounds (OTUs 11, 34, 26), degradation of dissolved organic matter (OTUs 4, 8, 16, 28, 33 and 40), and degradation of cellulose fibres (OTUs 32 and 9). Three of the bacterial OTUs are able to produce bioactive exometabolites (OTUs 30, 35 and 39). Results also revealed two pathogenic members, one belonging to Epsilonproteobacteria which can cause diarrhoea and septicaemia in humans (OTU38), and the other belonging to Flavobacterium which is known to be a causative agent to fish diseases (OTU24).
The most common 454 OTUs and ARISA OTUs increase in abundance as water moves downstream (into 3rd and 4th order streams). A few bacterial OTUs are more common in 1st order streams, apparently because they are terrestrial in origin, although site SC is different in respect of this trend given the influence of local septic waste discharges.
The 20 commonest ARISA OTUs make up about 40.4% of the total, while the 40 commonest 454 OTUs make up about 42% of all individuals, indicating that ARISA gives higher abundance but less diversity than 454 OTUs. So, either the same ARISA OTU can represent more than one 454 OTU or ARISA is overloading many OTUs because their abundance is below the limit of fluorescence (weight. off).
164 6.1.1 Comment on the techniques used in this research
Many studies have shown epifluorescence microscopy and DAPI staining to be reliable, successful and suitable direct methods for enumerating bacterial cells in freshwater (Hobbie et al. 1977; Porter and Feig 1980; Clarke and Joint 1986; Gasol et al. 1999; Garren and Azam 2010; Yamaguchi et al. 2011). Standard methods of determining heterotrophic plate counts using R2A medium and spread plates (Reasoner and Geldreich 1985) are efficient in recovering large numbers of bacteria from the targeted environments.
ARISA has been shown to provide reliable, robust and reproducible results of bacterial communities in the targeted environments (Fisher and Triplett 1999; Brown and Fuhrman 2005; Yannarell and Triplett 2005). It is considered to be a powerful technique that can be utilized to determine spatial and temporal shifts of bacterial community composition (Jones et al. 2007). Large samples can be processed by the ARISA tool and estimates of relative abundance of bacterial groups can be provided by it (Crump et al. 2003; Bending et al. 2007).
ITSF/ITSReub primer pair has been shown to be reproducible and can give a high number of peaks and wide spacer sizes (Cardinale et al. 2004).
Studies of bacterial communities have shown 454 pyrosequencing to be powerful tool that can give rapid characterization, better representation and large sequence depth of the bacterial community composition. It can be applied directly to environmental samples without the need for cloning (Edwards et al. 2006; Binladen et al. 2007; Bowers et al. 2009; Matcher et al. 2011). According to Engelbrektson (2010), amplicons of the hypervariable region (V1-V2) of the gene 16S rRNA was found to give higher estimates of the richness of bacterial species. In addition, because this study used error correcting barcodes, the assignment of sequences to samples was easy and successful (Parameswaran et al. 2007; Hamady et al. 2008). Also, according to Amend (2010), 454 pyrosequencing reads expressed the relative abundance of each taxon and can be used to compare between bacterial communities present in samples.